Fix line endings handling by BestMatchFontSelectorStrategy

DEVSIX-8688 Autoported commit. Original commit hash: [d7b728845]
aleks-ivanov · Oct 25, 2024 · 4d88123 · 4d88123
1 parent 2d0c66a
commit 4d88123
Show file tree

Hide file tree

Showing 13 changed files with 51 additions and 22 deletions.
diff --git a/itext.tests/itext.layout.tests/itext/layout/HyphenateLayoutTest.cs b/itext.tests/itext.layout.tests/itext/layout/HyphenateLayoutTest.cs
@@ -244,5 +244,20 @@ public virtual void HyphenSymbolTest01() {
             NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(outFileName, cmpFileName, destinationFolder
                 , "diff"));
         }
+
+        [NUnit.Framework.Test]
+        public virtual void WordsBreakingWordSoftHyphenTest() {
+            String outFileName = destinationFolder + "wordsBreakingWordSoftHyphenTest.pdf";
+            String cmpFileName = sourceFolder + "cmp_wordsBreakingWordSoftHyphenTest.pdf";
+            String SOFT_HYPHEN = "\u00AD";
+            String text = "Soft hyphen at the mid" + SOFT_HYPHEN + "dle,\nhyphen at the end: abcdef" + SOFT_HYPHEN + "ghijklmnopqrst\n"
+                 + SOFT_HYPHEN + "hyphen at the beginning.";
+            using (Document document = new Document(new PdfDocument(new PdfWriter(outFileName)))) {
+                document.Add(new Paragraph(text).SetWidth(150).SetBorder(new SolidBorder(1)).SetHyphenation(new HyphenationConfig
+                    (1, 1)));
+            }
+            NUnit.Framework.Assert.IsNull(new CompareTool().CompareByContent(outFileName, cmpFileName, destinationFolder
+                ));
+        }
     }
 }
diff --git a/...text.layout.tests/itext/layout/font/selectorstrategy/BestMatchFontSelectorStrategyTest.cs b/...text.layout.tests/itext/layout/font/selectorstrategy/BestMatchFontSelectorStrategyTest.cs
@@ -36,8 +36,8 @@ public virtual void TwoDiacriticsInRowTest() {
             IList<Tuple2<GlyphLine, PdfFont>> result = strategy.GetGlyphLines("L with accent: \u004f\u0301\u0302 abc");
             NUnit.Framework.Assert.AreEqual(3, result.Count);
             NUnit.Framework.Assert.AreEqual("L with accent: ", result[0].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual("\u004f\u0301\u0302", result[1].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual(" abc", result[2].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("\u004f\u0301\u0302 ", result[1].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("abc", result[2].GetFirst().ToString());
             // Diacritics and symbol were separated, but the font is the same
             NUnit.Framework.Assert.AreEqual(result[0].GetSecond(), result[2].GetSecond());
         }
@@ -49,8 +49,8 @@ public virtual void OneDiacriticTest() {
             IList<Tuple2<GlyphLine, PdfFont>> result = strategy.GetGlyphLines("L with accent: \u004f\u0302 abc");
             NUnit.Framework.Assert.AreEqual(3, result.Count);
             NUnit.Framework.Assert.AreEqual("L with accent: ", result[0].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual("\u004f\u0302", result[1].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual(" abc", result[2].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("\u004f\u0302 ", result[1].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("abc", result[2].GetFirst().ToString());
             NUnit.Framework.Assert.AreNotEqual(result[0].GetSecond(), result[1].GetSecond());
         }
 
@@ -75,8 +75,8 @@ public virtual void DiacriticFontDoesnotContainPreviousSymbolTest() {
             NUnit.Framework.Assert.AreEqual(6, result.Count);
             NUnit.Framework.Assert.AreEqual("Ми", result[0].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("\u0301", result[1].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual("ръ", result[2].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual(" (mír", result[3].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("ръ (", result[2].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("mír", result[3].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("ə", result[4].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual(")", result[5].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual(result[0].GetSecond(), result[2].GetSecond());
@@ -100,8 +100,8 @@ public virtual void SurrogatePairsTest() {
             IList<Tuple2<GlyphLine, PdfFont>> result = strategy.GetGlyphLines("text \uD800\uDF10\uD800\uDF00\uD800\uDF11 text"
                 );
             NUnit.Framework.Assert.AreEqual(3, result.Count);
-            NUnit.Framework.Assert.AreEqual("text", result[0].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual(" \uD800\uDF10\uD800\uDF00\uD800\uDF11 ", result[1].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("text ", result[0].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("\uD800\uDF10\uD800\uDF00\uD800\uDF11 ", result[1].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("text", result[2].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual(result[0].GetSecond(), result[2].GetSecond());
         }
@@ -122,13 +122,21 @@ public virtual void ThreeFontWithSpacesTest() {
             IFontSelectorStrategy strategy = FontSelectorTestsUtil.CreateStrategyWithLimitedThreeFonts(new BestMatchFontSelectorStrategy.BestMatchFontSelectorStrategyFactory
                 ());
             IList<Tuple2<GlyphLine, PdfFont>> result = strategy.GetGlyphLines(" axadefa ");
-            NUnit.Framework.Assert.AreEqual(6, result.Count);
+            NUnit.Framework.Assert.AreEqual(5, result.Count);
             NUnit.Framework.Assert.AreEqual(" a", result[0].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("x", result[1].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("a", result[2].GetFirst().ToString());
             NUnit.Framework.Assert.AreEqual("def", result[3].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual("a", result[4].GetFirst().ToString());
-            NUnit.Framework.Assert.AreEqual(" ", result[5].GetFirst().ToString());
+            NUnit.Framework.Assert.AreEqual("a ", result[4].GetFirst().ToString());
+        }
+
+        [NUnit.Framework.Test]
+        public virtual void WindowsLineEndingsTest() {
+            IFontSelectorStrategy strategy = FontSelectorTestsUtil.CreateStrategyWithFreeSans(new BestMatchFontSelectorStrategy.BestMatchFontSelectorStrategyFactory
+                ());
+            IList<Tuple2<GlyphLine, PdfFont>> result = strategy.GetGlyphLines("Hello\r\n   World!\r\n ");
+            NUnit.Framework.Assert.AreEqual(1, result.Count);
+            NUnit.Framework.Assert.AreEqual("Hello\r\n   World!\r\n ", result[0].GetFirst().ToString());
         }
     }
 }
diff --git a/.../resources/itext/layout/FontSelectorTest/cmp_checkThreeFontsInOneLineWithUnicodeRange.pdf b/.../resources/itext/layout/FontSelectorTest/cmp_checkThreeFontsInOneLineWithUnicodeRange.pdf
diff --git a/.../itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup.pdf b/.../itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup.pdf
diff --git a/...itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup2.pdf b/...itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup2.pdf
diff --git a/...itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup3.pdf b/...itext.layout.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinGroup3.pdf
diff --git a/...ut.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinWithUnicodeRange.pdf b/...ut.tests/resources/itext/layout/FontSelectorTest/cmp_cyrillicAndLatinWithUnicodeRange.pdf
diff --git a/...ayout.tests/resources/itext/layout/FontSelectorTest/cmp_duplicateFontWithUnicodeRange.pdf b/...ayout.tests/resources/itext/layout/FontSelectorTest/cmp_duplicateFontWithUnicodeRange.pdf
diff --git a/....tests/resources/itext/layout/HyphenateLayoutTest/cmp_wordsBreakingWordSoftHyphenTest.pdf b/....tests/resources/itext/layout/HyphenateLayoutTest/cmp_wordsBreakingWordSoftHyphenTest.pdf
diff --git a/...ests/itext.layout.tests/resources/itext/layout/LayoutTaggingTest/cmp_notAsciiCharTest.pdf b/...ests/itext.layout.tests/resources/itext/layout/LayoutTaggingTest/cmp_notAsciiCharTest.pdf
diff --git a/itext/itext.layout/itext/layout/font/selectorstrategy/AbstractFontSelectorStrategy.cs b/itext/itext.layout/itext/layout/font/selectorstrategy/AbstractFontSelectorStrategy.cs
@@ -109,7 +109,8 @@ public virtual IList<Tuple2<GlyphLine, PdfFont>> GetGlyphLines(String text) {
                         if (codePoint > 0xFFFF) {
                             i++;
                         }
-                        if (IsCurrentFontCheckRequired() && (i != indexDiacritic - 1)) {
+                        if (IsCurrentFontCheckRequired() && (i != indexDiacritic - 1) && !iText.IO.Util.TextUtil.IsWhitespaceOrNonPrintable
+                            (codePoint)) {
                             if (currentFont != MatchFont(codePoint, fontSelector, fontProvider, additionalFonts)) {
                                 breakRequested = true;
                             }
@@ -191,8 +192,7 @@ protected internal virtual PdfFont MatchFont(int codePoint, FontSelector fontSel
             foreach (FontInfo fontInfo in fontSelector.GetFonts()) {
                 if (fontInfo.GetFontUnicodeRange().Contains(codePoint)) {
                     PdfFont temptFont = GetPdfFont(fontInfo, fontProvider, additionalFonts);
-                    Glyph glyph = temptFont.GetGlyph(codePoint);
-                    if (null != glyph && 0 != glyph.GetCode()) {
+                    if (temptFont.ContainsGlyph(codePoint)) {
                         matchedFont = temptFont;
                         break;
                     }

diff --git a/itext/itext.layout/itext/layout/renderer/TextRenderer.cs b/itext/itext.layout/itext/layout/renderer/TextRenderer.cs
@@ -466,20 +466,26 @@ public override LayoutResult Layout(LayoutContext layoutContext) {
                                         for (int i = hyph.Length() - 1; i >= 0; i--) {
                                             String pre = hyph.GetPreHyphenText(i);
                                             String pos = hyph.GetPostHyphenText(i);
-                                            float currentHyphenationChoicePreTextWidth = GetGlyphLineWidth(ConvertToGlyphLine(text.ToUnicodeString(currentTextPos
-                                                , wordBounds[0]) + pre + hyphenationConfig.GetHyphenSymbol()), fontSize.GetValue(), hScale, characterSpacing
-                                                , wordSpacing);
+                                            char hyphen = hyphenationConfig.GetHyphenSymbol();
+                                            String glyphLine = text.ToUnicodeString(currentTextPos, wordBounds[0]) + pre;
+                                            if (font.ContainsGlyph(hyphen)) {
+                                                glyphLine += hyphen;
+                                            }
+                                            float currentHyphenationChoicePreTextWidth = GetGlyphLineWidth(ConvertToGlyphLine(glyphLine), fontSize.GetValue
+                                                (), hScale, characterSpacing, wordSpacing);
                                             if (currentLineWidth + currentHyphenationChoicePreTextWidth + italicSkewAddition + boldSimulationAddition 
                                                 <= layoutBox.GetWidth()) {
                                                 hyphenationApplied = true;
                                                 if (line.GetStart() == -1) {
                                                     line.SetStart(currentTextPos);
                                                 }
                                                 line.SetEnd(Math.Max(line.GetEnd(), wordBounds[0] + pre.Length));
-                                                GlyphLine lineCopy = line.Copy(line.GetStart(), line.GetEnd());
-                                                lineCopy.Add(font.GetGlyph(hyphenationConfig.GetHyphenSymbol()));
-                                                lineCopy.SetEnd(lineCopy.GetEnd() + 1);
-                                                line = lineCopy;
+                                                if (font.ContainsGlyph(hyphen)) {
+                                                    GlyphLine lineCopy = line.Copy(line.GetStart(), line.GetEnd());
+                                                    lineCopy.Add(font.GetGlyph(hyphen));
+                                                    lineCopy.SetEnd(lineCopy.GetEnd() + 1);
+                                                    line = lineCopy;
+                                                }
                                                 // TODO DEVSIX-7010 recalculate line properties in case of word hyphenation.
                                                 // These values are based on whole word. Recalculate properly based on hyphenated part.
                                                 currentLineAscender = Math.Max(currentLineAscender, nonBreakablePartMaxAscender);

diff --git a/port-hash b/port-hash
@@ -1 +1 @@
-238107cd5532d937fc246b97f60c8f1599786f9d
+d7b7288459c53470547f639cf5386cb2636c2eca